YUZO HOSHINO 279 
In the above table, we meet with so large a number of whites in the late 
flowering- group, tliat, when we t;ike it as the unit and calculate ratios, the 
result differs much from the expected. By considering, however, that on 
account of environmental influences the retarding of the flowering time is of 
common occurrence and that some individuals belonging to the early group 
genetically may very often fall thereby into the late group, we may take the 
number of reds in the early group as the unit and calculate ratios. Then we 
get the following ratios : — 
Early group Late group 
W\ R. R. 
Observed 1.3 i.o 0,45 3.8 
Expected 1.2 i.o 0.26 3.3 
Here, except whites in the late group, the ratios are quite close to the expected. 
Thus, our propo.sed gametic coupling is practically confirmed, but before 
we draw a definite conclusion, we have to prove the following two proposi- 
tions : — 
1. Non-presence of the correlation in Eg and F^^ variable families, which 
are the descendants of the E^ plants which have become homozygous in 
the factor A or a (aabB or AAbB). 
2. The occurrence of the zygotic series 49 : 15 : 15 : 177 in Eg and E^. vari- 
able families which are the descendants of the plants heterozygous 
only in the factor A (aAbb or aABB). 
(i) As we have already noted on page 259, the distinction between the 
supposed aabB families and the supposed aAbb families in E^ raisings (Tables 
9, 10 and 11) is not clear, but by careful examination of the variable families 
descended from the early-flowered Eg plants, we see that those families which 
have relatively narrow ranges of variation and which suggest the progenies 
of aabB parents, have uniform distribution of white-flowered individuals and 
indicate non-presence of the correlation (families, Nos. 26, 27 and 28 in Table 
9 ; Nos. 19, 20, 21 and 22 in Table 10 ; No. 20 in Table 1 1 ). In Table 14 H, 
where the results of the raising of E^^ progenies of the family No. 20 Table. 
